基于小波变换和BP神经网络的视觉诱发电位识别

The Recognition of Visual Evoked Potential Based on Wavelet and BP Neural Network

结合小波变换和BP神经网络对视觉诱发脑电信号(VEP)进行分类而产生脑机接口控制信号。利用一维离散小波变换提取强噪声背景下的低频微弱脑电信号,获取特征向量输入BP神经网络进行事件相关电位模式识别。实验表明,小波变换特征向量提取方法能有效地实现信号的去噪、降维和特征提取,BP神经网络能比较准确地从VEP中识别出事件相关电位,进行10次测试的平均识别正确率为99.375%,有利于产生脑机接口控制信号。

The classification of visual evoked potential（VEP） with wavelet transform and BP neural network for obtaining brain-computer interface （BCI） control signal. One-dimensional discrete wavelet transform （DWT） method are used to extract the low-frequency weak VEP signal from strong background noise.The extracted feature vectors are input to the BP neural network to achieve the recognition of event related potential（ERP）. Experiments show that the feature extraction in wavelet transformation can effectively extract VEP features, reduce noise and decrease the dimensionality and the BP neural network can recognize ERP from VEP signal correctly, 10 tests, the average correct identification rate of 99.375%,which is useful to produce BCI control signal.